Research Of Automatic Transfer Function Based On Geological Target Drive

Volume rendering is an important three-dimensional data visualization method. The basic idea is the use of a transfer function that maps the voxel values into RGBA optical parameters. By adjusting the value of A, users can achieve the effect of perspective. Since the volume rendering visualization is closely related to the transfer function, so the transfer function design has always been important and difficult in the field of research. Transfer function essentially reflects the characteristics of the target. When the target feature is missing or unknown, the design of transfer function can’t be accurately. Traditional transfer function design method is called trial and error. Traditional method must adjust the transfer function continuously. The results will stay updated, until it meets the users’demand. In this thesis, in order to reduce the difficulty of the transfer function design and improve design efficiency. We use the signal processing and pattern recognition methods and ideas, to carry out the transfer function of the automatic design method.For low signal to noise ratio of seismic data, geological targets showing the layered distribution and lack of obvious boundary feature basic, this thesis carried out the function design method based on geological target-driven automatic transfer. The main contributions are as follows:1. We proposed a transfer function designing method based on density and linear related. This method mainly overcome the disadvantage of seismic waveform data based on single-channel. Since the layer points site at peak or trough position, so we can show the characteristics of layered geological targets by extracting the sampling point at peak or trough position. But there are obvious shortcomings method based on single-channel waveform:When noise points are at peak or trough, the method can’t distinguish normal horizon and noise. In this paper we use multi-channel seismic data and proposed the transfer function based on the density by utilizing the characteristics of the discrete sparsely distributed. The dense of noise is less than horizon; By utilizing horizontally adjacent data correlation strong features, we presented linear correlation coefficient based on the noise elimination method, that horizon point and surrounding property is similar to the sampling point. And noise points often unexpected presence, compared to the surrounding points. Finally the voxel values will be automatically converted to transparency2. We proposed a method of transfer function designing based on scale space. The method is mainly for this situation when noise get together like cluster and the noise points’ density is similar to the layer points’. It is easy to form a pseudo horizon structure. Pseudo-horizon structure density may be similar to real horizon’s, but it is significantly smaller in the space occupied by the scale. We use spot detection technology to determine the block size, and then distinguish layers and noise based on size scale. Finally this method can convert the voxel values a to transparency automatically.3. We design and implement GPU-based rendering body simulation platform. The platform uses several transfer function method proposed in this paper when processes drawing. And we provide practical verification of theoretical paper.In summary, to overcome the problem about geological targets-driven automatic transfer function design, we proposed effective solutions to reduce the complexity of transfer function design, and clearly shows the layered geological characteristics.